The primary antibodies used for flow cytometry, immunofluorescence, immunohistochemistry, and immunoblotting were CD206 (Cat. no. 321104; BioLegend), CD163 (Cat. no. 333606; BioLegend), α-SMA (Cat. no. ab5694, Abcam,), and GAPDH (Cat#: 5174; Cell Signaling Technology). CD14 microbeads (Cat. no. 130-050-201, Miltenyi Biotec) were used to sort monocytes. Butylated hydroxyanisole (BHA) was purchased from Sigma–Aldrich.

Human pancreatic CAF-Stellate Cells were purchased from Neuromics and cultured in CAF growth media (Neuromics). BxPC3_Red_luc cells were cultured in RPMI1640 or Dulbecco’s Minimal Essential Medium supplemented with 10% fetal bovine serum.

For 2D cultures, 5x10⁵ human pancreatic CAFs or a mix of 1x10⁵ BxPC3 (RRID:CVCL_XX78) cancer cells and 3x10⁵ pancreatic CAFs per well were cultured in 6-well plates in 2 ml RPMI 1640 with 0.5% fetal bovine serum (FBS) culture media for monoculture and co-culture, respectively. For 3D culture, 2x104 pancreatic CAFs or a mix of a mix of 0.5x10⁴ BxPC3 cells with 1.5x10⁴ pancreatic CAFs per well were seeded in 100 μL of RPMI 1640 with 0.5% FBS culture medium in 96-well round-bottom, ultra-low attachment plates (Corning Inc., New York, USA), centrifuged at 500×g for five minutes then placed in a 37°C incubator at 5% CO2 for 2 days. The cells and spheroids were harvested, washed with DPBS, and frozen for RNA-seq analysis. Supernatants were collected and centrifuged to remove cell debris for Olink analysis. The quality control (QC) for single-nucleus RNA sequencing (snRNAseq) data analysis is provided in S1 Fig.

5x10³ BxPC3 mCherry-luficerase reporter cells were monocultured or co-cultured with 1x10⁴ CAFs in flat-bottom or ultra-low attachment 96-well plates for 2D and 3D cultures, respectively. The Incucyte live-cell imaging platform (Sartorius) tracked BxPC3 killing over a period of 5 days for both 2D and 3D cultures. At the end of the imaging, the BxPC3 residual cell count was assessed using the Bright-Glo luciferase system (Cat. No. E2620, Promega) on the GloMax plate reader (Promega).

Primary peripheral blood human NK cells were acquired from healthy nonsmoking donors (Cat. no.70036, Stem Cell Technologies). NK cells were cultured in CTS NK Xpander medium (Cat. No. A5019001, ThermoFisher Scientific), supplemented with 5% human AB serum (Cat. No.H4522, Sigma Aldrich), and 200 ng/ml IL-2 (Cat. no. 202-IL-010; R and D Biosystems) for 48 h post thaw to generate activated NK cells.

For the NK cell killing assay, IL2 activated NK cells were cultured with either BxPC3 pancreatic adenocarcinoma cells or co-cultured BxPC3 and CAF cells in both monolayer 2D and spheroid 3D format in 96 well tissue culture plates (Cat. No. 3595 and 4520, Corning), as described in the section on tumor growth measurements. Effector (NK cells) to target (BxPC3) ratios of 5:1, 2.5:1, 1.25:1, and 0.625:1 were used in the NK cell killing assay. The Incucyte live-cell imaging platform (Sartorius) tracked BxPC3 killing over a period of 7 days for both 2D and 3D cultures. At the end of the imaging, the BxPC3 residual cell count was assessed using the Bright-Glo luciferase system (Cat. No. E2620, Promega) on the GloMax plate reader (Promega).

Human pancreatic CAFs were cultured in the CAF growth medium (Cat. No. CAFM03, Neuromics). A total CAF cell density of 0.335 × 106 cells/ml was maintained for both monolayer (2D) and spheroid (3D) CAF culture. A total of 1.34 x 106 cells were cultured in 4 ml reduced serum medium (RPMI-1640, 2 mM glutamine, and 0.5% Fetal Calf Serum) in each well of Aggrewell 800 (Cat. no. 34825; Stem Cell Technologies) for 3D culture with 900 cells seeded per microcavity. The 6-well Aggrewell microcavity plate was briefly centrifuged at 300 × g for 2 min to allow CAF spheroid formation. For 2D CAF culture, 0.668 × 106 cells were cultured in 2 ml reduced serum medium in each well of a 6 well tissue culture plate (Cat. No. 140675, Nunc Delta, ThermoFisher Scientific). Conditioned supernatant was collected 72 h post CAF culture for both 2D and 3D, filtered through a 40-u cell strainer, and frozen at -80°C.

Leukocytes were obtained from healthy volunteers. Monocytes were isolated using CD14 negative selection methods by Robosep from StemCell. The purity and recovery of CD14+ monocytes were determined using CD14-FITC antibody labeling (Cat. no. 130-113-708; Miltenyi Biotec), whereas cell viability was determined by propidium iodide (PI) staining and flow cytometry analysis using a BD FACSAria III (BD Biosciences, San Jose, CA, USA).

Human endothelial cells (HUVECs, RRID:CVCL_2959) were purchased from StemCells and cultured in EBM^TM-2 Endothelial Cell Growth Basal Medium-2 supplemented with 5% fetal bovine serum (FBS), hydrocortisone, human fibroblast growth factor-beta (hFGF-b), vascular Endothelial Growth Factor (VEGF), R3-Insulin-like growth factor-1 (R3-IGF-1), ascorbic acid, epidermal growth factor (hEGF), and gentamicin/amphotericin-B (GA). IncuCyte ClearView 96-Well Chemotaxis Cell Migration Plates were coated with 5 mg/mL fibronectin diluted with D-PBS, and 8000 HUVECs per well were seeded into the top chamber of a Transwell and incubated overnight to form a confluent monolayer. Subsequently, monocytes pre-stained with the green CellTracker dye were added to the top chamber. Monocyte migration was monitored by imaging the top and bottom membranes of cells in an incubator. Quantitative data demonstrated that conditioned media from 3D CAFs significantly enhanced monocyte transendothelial migration compared to that from 2D CAFs. Fluorescent labeling of monocytes and chemotaxis plate preparation with HUVEC monolayers. Real-time analysis using the IncuCyte® S3 live-cell imaging system, image analysis, and assessment of the transendothelial migration rates. CCL2 (100 ng/ml) was used as a positive control.

Monocytes were isolated from the leukopaks of healthy non-smoking donors (cat. no. 70500.2, Stem Cell Technologies), using the Easy Sep Human Monocyte Isolation Kit (Cat. 19356 RF, Stem Cell Technologies) on a RoboSep cell isolation automated platform (Stem Cell Technologies), according to the manufacturer’s instructions. The purity of the isolated monocytes was verified using flow cytometry (CD14+ CD15 −). Monocytes were cryopreserved in Cryostor CS10 (Cat. no. 210502; Biolife Solutions).

Frozen monocytes were thawed and cultured in Immunocult SF Macrophage Medium (Cat. no. 10961, Stem Cell Technologies). A total of 0.5 × 106 cells were seeded in each well of 12 well tissue culture treated plate in 500 μL medium. Conditioned medium (500 μL) from the 2D or 3D CAF cultures was added to 50% of the total volume. The effect of CAF-conditioned medium on macrophage differentiation and polarization was assessed 48-h and 7 d after CAF-conditioned medium treatment. On days 2 and 7, cells were harvested using Accutase (Cat. No. 00-4555-56, Thermo Fisher Scientific) and analyzed using flow cytometry.

Macrophages were washed with 1X DPBS and stained with fixable aqua (Cat. No. L34957) at 1:1000 dilution in 1X DPBS for 15 min. The cells were washed with FCS staining buffer (Cat. N0.554647; BD Biosciences), and incubated with an Fc receptor blocker (Cat. No.422302; BioLegend) for 10 min. Primary conjugated antibodies (S1 Table) were incubated for 20 min in the dark at room temperature. IgG controls were included in all antibodies (S2 Table). Following incubation with the primary antibody, cells were washed with FCS buffer. The cells were incubated for 30 min at 4 C in the dark. The cells were then washed twice with FCS buffer (800 g, 2 min). For intracellular CD68 staining, 100 μL fixation/permeabilization solution was added to the cells for 20 min. Next, the cells were washed twice with 1× BD Perm/Wash Buffer. CD68 antibody was added at a 1:100 dilution in Perm/Wash buffer, and the cells were incubated at RT for 30 min in the dark. Cells were washed twice in perm/wash buffer and resuspended in Staining FCS Buffer prior to flow cytometric analysis. The cells were analyzed using an Attune Nxt flow cytometer (ThermoFisher Scientific). The gating strategies employed for the FACS analysis of macrophage phenotypes were as follows: dead cells were excluded from the analysis; monocyte-derived macrophages were gated as M1 (CD45+ CD14 lo/- CD11b+ CD80high/+ CD86 high/+), M2a (CD45+ CD14+ CD11b+ CD206 high/+), and M2c (CD45 + CD14+ CD11b+ CD163 high/+).

The panels of macrophage-specific markers and IgG controls for flow cytometry analysis are listed in S1 and S2 Tables respectively.

CAF-secreted protein expression in the cell-conditioned medium was analyzed using the MSD multi-array multiplex platform, according to the manufacturer’s instructions. (Meso Scale Discovery, Rockville, MD, USA). Briefly, MSD multi-array 96 well plates were coated with a spot-specific linker and capture antibody mix, and incubated with shaking for 60 min at 750 rpm at room temperature. After capture antibody incubation, CAF-conditioned medium samples were added to the plates at 1:2 and 1:10 dilutions along with the standard calibrators provided in the kit. The Samples and calibrator standards were incubated overnight at 4°C. Following overnight incubation, sulfo-tag-conjugated detection antibodies were added to the samples and incubated at room temperature with shaking at 750 rpm for 60 minutes. The plate was washed thrice with MSD buffer between each step. The readout was performed after the addition of MSD Gold Read Buffer. ECL was detected using an MSD Meso Sector S600 MM reader and the data were analyzed using the Discovery Workbench.

This article does not contain any studies involving human participants performed by any of the authors.

Pancreatic ductal adenocarcinoma (PDAC) is characterized by a significant amount of desmoplasia, accounting for up to 90% of the total tumor volume [30, 31]. The microenvironment of pancreatic ductal adenocarcinoma (PDAC) is distinguished by a dense, fibrotic stroma, with cancer-associated fibroblasts (CAFs) serving as pivotal contributors [32]. Recently, researchers have classified various subtypes of CAFs in PDAC patients according to their transcriptome signatures [33, 34]. To understand how the growth of PDAC patient-derived primary CAF under 2D and 3D conditions influences the transcriptomic profiles of both CAFs and cancer cells, we used single-nucleus RNA sequencing (snRNA-seq) to detect highly sensitive and cell-specific gene expression. Simultaneously, we collected the culture supernatants and assessed the expression of a panel of immuno-oncology and inflammation markers by using a proximity extension assay (Fig 1A).

The human pancreatic cancer cell line BxPC3 and primary patient-derived pancreatic CAFs were used for TME modeling studies. Clustering analysis of snRNA-seq data confirmed that CAFs and cancer cells exhibited distinct transcriptional profiles based on the dimensionality of their growth (Fig 1B). Despite the lack of a consensus CAF molecular signature, several biomarkers have been used to distinguish different CAF subtypes in PDAC patient samples, including ACTA2, TAGLN, MYL9, TPM1, TPM2, HOPX, and POSTN for myofibroblastic CAFs (myCAFs), and CXCL8, CXCL2, PDGFRA, IL6, CFD, DPT, AGTR1, HAS1, CXCL1, CCL2, IL8, and LMNA for inflammatory CAFs (iCAFs) [35]. Interestingly, CAFs cultured in 2D, whether in monoculture or co-culture with tumor cells, exhibited elevated expression levels of myCAF-associated marker genes (Fig 1C). Conversely, CAFs cultured in 3D monoculture showed increased expression of marker genes associated with iCAFs, including CCL2, CXCL8, and PDGFRA, but had indistinguishable IL6 levels compared to CAFs cultured in 2D and 3D. Interestingly, iCAF signatures were subdued on CAFs when co-cultured with cancer cells in a 3D culture. We did not observe any antigen-presenting CAFs (apCAFs) based on the expression markers H2-Aa, H2-Ab1, CD74, CD239, or CD321 in any 2D or 3D CAF model.

The transcript levels detected by snRNA-seq and the secreted protein levels produced by CAFs under varying culture conditions showed significant compositional differences (Fig 2D). For the most part, gene transcript levels measured by snRNA-seq and protein levels measured by Olink for the same gene were strongly correlated, with a few exceptions (Fig 2B), implying that transcriptomic values reliably represented translated protein levels. Notably, 3D-cultured CAFs exhibited elevated transcript and protein expression levels of CCL7, CCL8, IL24, LIF, VEGFA, CXCL8, TNFRSF11B, and MMP10 compared to their 2D counterparts. These genes play critical roles in biological pathways related to inflammation, immune responses, tissue repair, and angiogenesis and emphasize the distinct gene expression differences of CAFs cultured in 2D versus 3D.

We observed distinct transcriptomic profiles in cancer cells and CAFs when co-cultured with CAFs under 2D and 3D conditions, as revealed by the snRNAseq data analysis (Fig 2A). Notably, CAFs in 2D monoculture showed elevated expression of TGFB2, NREP, MEGF6, and POU2F2, whereas CAFs in 2D co-culture showed elevated expression of PTN, ALDH1A1, PGS10, MFAP4, RGCC, LHB, ALDH2 and LMCD1. CAFs grown in 3D monoculture showed elevated expression of CCL7, LC16A6, NR4A2, MMP3, BDKRB2, CHI3L1, TWIST1, GK, HGH, IL24, and TRPA1, whereas CAFs in 3D co-culture showed elevated expression of FAM2168, LINC01705, MMP11, and MMP9 (Figs 2B and 3C).

The role of CAFs in promoting tumor progression has been extensively reported in different types of solid organ cancers [36, 37]. Because CAFs cultured in 2D or 3D displayed different transcriptomic/proteomic profiles of CAF subtypes, we next assessed how these subtypes affected cancer cell growth. The BxPC3 PDAC cell line was engineered to express both a red fluorescent protein and luciferase reporter, allowing us to monitor cancer cell growth using both high-content imaging (Incucyte) and luciferase assays (Brite-Glo) (Fig 2A). Fig 3B and 3C display images of red fluorescent cancer cells in both monoculture and co-culture with CAFs in 2D and 3D and reveal significant growth differences in 3D compared to 2D. We tracked the number of cancer cells in 2D cultures over time and noticed that co-culture with CAFs in a 2D environment did not significantly affect cancer cell growth, as evidenced by both the imaging analysis and endpoint luciferase assays (Fig 3D and 3E). In the 3D setting, we assessed the overall fluorescence intensity of tumor spheroids with or without CAFs and observed that co-culture with CAFs in 3D yielded a substantial increase in cancer cell growth compared to cancer cell monoculture in 3D (Fig 3F). Quantification of cancer cell growth by luciferase assay confirmed that co-culture with CAFs in 3D culture resulted in significantly greater tumor spheroid growth than 3D cancer cell monoculture (Fig 3G). These data revealed protumorigenic properties of CAFs grown in 3D co-cultures that were not observed when cultured in 2D formats.

CAFs play a pivotal role in solid tumor growth and maintenance by actively remodeling the stromal ECM through the secretion of matrix metalloproteinases (MMPs) [38, 39]. This remodeling process provides essential structural support for tumor invasion and neoangiogenesis. We dissected the impact of culture dimensions on MMP expression by using transcriptomic and proteomic data. Notably, CAFs grown in the 3D monoculture exhibited elevated expression levels of MMP1, MMP2, MMP3, MMP10, and MMP14 compared to their 2D counterparts (Fig 3B). The transcriptomic data aligned with the equivalent protein expression levels detected in the supernatants from 3D co-cultures, which exhibited increased levels of MMP1, MMP7, MMP9, MMP10, and MMP12 compared to 2D co-cultures (Fig 3C). Interestingly, the elevated expression of MMP1, MMP2, and MMP3 observed in 3D CAF monocultures was downregulated in 3D co-cultures, indicating a tumor-mediated suppressive effect on these genes. Intriguingly, MMP9 expression was specifically upregulated in both tumor cells and CAFs only in the 3D co-cultures [40].

Proteomic analyses revealed similarly distinct expression levels of secreted proteins in the supernatants of CAFs and cancer cell co-cultures (Fig 3C). Notably, several cytokines of the CXCL family displayed higher expression levels in 2D than in 3D co-cultures. These proteins function in immune cell recruitment, tumor biology, and disease pathogenesis.

Natural killer (NK) cells play an essential role in tumor immune surveillance, exhibiting an innate ability to kill tumor cells without prior priming [41, 42]. To understand the functional implications of divergent gene and protein expression in CAFs grown in different formats, we explored the effect of CAFs on tumor sensitivity to activated human primary NK cells. We used the PDAC TME model, BxPC3 cells that express a red fluorescent protein and luciferase reporter, to enable both high-content kinetic imaging (Incucyte) and tumor viability (Brite-Glo) assays to monitor cancer cell growth (Fig 4A). We assessed NK cell-mediated antitumor function at varying effector-to-target (E:T) ratios to identify dose-responsive effects. High-content kinetic imaging studies revealed substantial NK-mediated tumor cell killing at all E:T ratios in 2D growth formats (Fig 4B). The data from both Incucyte imaging and luciferase end-point assays were consistent with 2D cell cultures (Fig 4C and 4D), and activated human primary NK cells effectively eradicated cancer cells regardless of the presence or absence of CAFs. However, these NK antitumor effects were significantly suppressed when cancer cells were grown in 3D and in the presence of CAFs (Fig 4E). We observed a modest reduction in the fluorescence intensity of cancer cells when the spheroids in the 3D co-culture were treated with activated NK cells compared to that of the 3D tumor monoculture (Fig 4F). This observation was confirmed by the end-point luciferase results (Fig 4G), which suggested that CAFs might confer protection to tumor cells from NK cell killing in 3D culture settings.

NK cell activation and effector functions rely on a delicate balance between activating and inhibitory cell surface ligand-receptor interactions independent of prior sensitization and MHC restriction [42, 43]. We investigated the expression of NK cell-interacting molecules on tumor cell surfaces co-cultured with CAFs in 2D and 3D formats. Notably, tumor cell co-culture with CAFs in 3D led to the downregulation of pro-tumor ligands (HLA-A, B, C, and E) in cancer cells compared to 2D co-culture (Fig 4H). On the other hand, expression of a pro-inflammatory molecule TNFRSF21 is considerably high in both 2D and 3D cultures. Interestingly, the expression levels of IL-15 and IL-18, which are cytokines critical for NK cell maturation, proliferation, and antitumor function, were higher in the 2D co-culture (Fig 4I). Unexpectedly, cytokines, including transforming growth factor TGFb, IL-1b, and IL-10, which can impede NK cell functions, were also upregulated in the 2D co-culture compared to the 3D co-culture.

Previous studies have highlighted the role of CAFs in modulating immune cell behavior [44, 45]. Specifically, CAFs have been implicated in the recruitment and transendothelial migration of monocytes from circulating vasculature to tumor sites through the secretion of chemokines (including CCL2) and induction of chemotactic gradients [37].

We investigated how conditioned media (CM) derived from CAFs cultured in both 2D and 3D environments affected monocyte transendothelial migration. Conditioned media that were collected from primary PDAC CAFs cultured in both 2D and 3D environments served as a source of cytokines. Human umbilical vein endothelial cells (HUVECs) were seeded onto the top chamber of a transwell and allowed to adhere overnight, forming a confluent monolayer of vascular endothelium. CD14+ monocytes isolated from healthy human donor leukopaks were stained with CellTracker Green CMFDA dye and subsequently added to the top chamber (Fig 5A). Monocyte migration was monitored using kinetic high-content imaging. The captured images illustrate the increased migration of monocytes from the top (blood vessel) to the bottom (tumor compartment) of the Transwell membrane during incubation (Fig 5B). Quantitation of imaging data showed that conditioned media from 3D CAFs considerably enhanced monocyte transendothelial migration compared to control media (Fig 5C). Conditioned media from 2D CAFs promoted only modest monocyte migration, not significantly more than control media (Fig 5C). These data revealed a secreted chemotactic component from 3D CAFs that contributes to monocyte recruitment and tumor infiltration, which is not secreted by 2D CAFs.

To better understand the secreted factors responsible for monocyte transendothelial migration, we examined the expression levels of key cytokines related to monocyte migration in 2D and 3D CAF cultures. While the transcript levels of CCL2 and MMP2, known enhancers of monocyte migration, were significantly higher in 3D cultured CAFs compared than in 2D cultures (Fig 5D), the protein expression levels of CCL2 (MCP-1) remained comparable in the supernatants from both 2D and 3D CAF monocultures (Fig 5E). However, our results revealed that several other cytokines associated with monocyte migration, such as IL-8, MCP2, MCP-3, CXCL-1, and CXCL-5, were significantly elevated in 3D CAF-conditioned media compared to their 2D counterparts (Fig 5E). Additionally, we performed an enzyme-linked immunosorbent assay (ELISA) to measure the expression of several selected cytokines including IL8, CCL4, CCL2, and CCL7, in the supernatants of 2D and 3D CAFs cultures. We confirmed that the expression levels of CCL2 in 2D and 3D conditioned media were comparable, and validated the increased expression levels of IL-8, CCL4, and CCL7 in 3D CAFs compared with those in 2D CAFs (Fig 5F). These data revealed a clinically relevant chemotactic mechanism [46] that induces monocyte transendothelial migration into tumors, which can only be reproduced using 3D CAF models.

Tumor-associated macrophages (TAMs) are strategically positioned near CAF-populated regions within the tumor stroma, where they exert protumorigenic functions [47]. These immune cells are crudely classified into two distinct subsets, M1 and M2 (where M1 is anti-tumorigenic and M2 is protumorigenic), each is activated by specific polarizing cytokines and plays distinct roles in cancer progression [48–50]. Given the propensity of 3D CAFs to induce monocyte transendothelial migration, we next investigated how they would affect subsequent monocyte differentiation and macrophage polarization once infiltrated into the TME.

We cultured monocytes in conditioned media derived from both 2D and 3D CAF cultures for 48 h and then used multicolor flow cytometry analysis to phenotype polarized macrophages based on distinct cell surface marker expression (Fig 6A). Monocytes grown in the presence of 3D CAF-conditioned media demonstrated elevated cell surface expression levels of CD206 and CD163, which are canonical protumoral markers of M2 TAMs, compared with those grown in basal media and 2D CAF-conditioned media (Fig 6B).

CAF-secreted cytokines play critical roles in monocyte recruitment and transformation into M2 macrophages [51, 52]. Specifically, GM-CSF, M-CSF, IL-4, IL-13, IL-8, IL-10, TGF-β, Il-1b, and CCL2 have been shown to promote M2 polarization [45]. Conversely, TNF-α, IFN-γ, TGF-β, VEGF, IL-6, and CXCL12 skew macrophage differentiation toward the M1 phenotype [53]. We quantified the expression levels of several key cytokines in the conditioned media of the 2D and 3D CAF cultures using ELISA. Notably, the levels of M2-inducing cytokines, such as IL-10, Il-1b, GM-CSF, and M-CSF, were significantly higher in 3D CAF-conditioned media than in 2D media (Fig 6C). Conversely, IL-6, an M1 inducing cytokine, was elevated in 2D CAF-conditioned media compared to 3D conditioned media (Fig 6C). Collectively, these observations indicate that CAFs grown in a 3D tumor spheroid format create chemotactic gradients that facilitate monocyte migration from the tumor vasculature and the subsequent polarization into protumorigenic M2 macrophages.